Skip to content

Instantly share code, notes, and snippets.

@Ngoguey42

Ngoguey42/.ocamlformat

Last active August 6, 2021 16:52
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save Ngoguey42/872ded9908593e1dc94ab0f28e8abbe8 to your computer and use it in GitHub Desktop.
Save Ngoguey42/872ded9908593e1dc94ab0f28e8abbe8 to your computer and use it in GitHub Desktop.
Download ZIP
irmin: bench/tree.exe bootstrap trace plot
Raw
dune
(env
(dev (flags (-w +1..3+5..28+30..39+43+46..47+49..57+61..62-40-26 -strict-sequence -strict-formats -short-paths -keep-locs -g)))
(release (flags (-w +1..3+5..28+30..39+43+46..47+49..57+61..62-40-26 -strict-sequence -strict-formats -short-paths -keep-locs )))
)
(executable
(name json_to_repr)
(libraries repr ppx_repr fmt ppx_deriving_yojson mtime mtime.clock.os unix re)
(preprocess
(pps ppx_repr ppx_deriving_yojson)))
Raw
dune-project
(lang dune 2.8)
Raw
histo_tree.py
"""Plot distribution of timings out of bench/tree.exe on boostrap trace
Formatted using YAPF
"""
import argparse
import json
import os
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib
import matplotlib.pyplot as plt
sns.set_theme(style="ticks", color_codes=True)
plt.rcParams["legend.labelspacing"] = 1.7
p = argparse.ArgumentParser()
p.add_argument('--in',
help="""path to JSON file.
Pseudo schema:
{
"revision" : string;
"flatten": int;
"inode_config": string;
"max_durations": {string -> (int, float)};
"moving_average_points": {string -> {"x": float array; "y": float array}};
"histo_points": {string -> (int * float) array};
}
""")
p.add_argument('--out', help="Suffix of output filename")
p.add_argument('-o',
help="Overwrite if --out exists. Fail otherwise",
action="store_true")
args = p.parse_args()
print(args)
j = json.loads(open(dict(args._get_kwargs())['in']).read())
if args.o == False and os.path.exists(args.out):
raise NameError("{} exists!! Pass -o to overrite it".format(args.out))
def time_to_txt(t):
h = np.floor(t / 3600)
t -= h * 3600
m = np.floor(t / 60)
t -= m * 60
return "{:02.0f}:{:02.0f}:{:02.0f}".format(h, m, t)
def time_to_txt2(t):
if t < 1e-6:
return '{:.0f} ns'.format(t / 1e-9)
if t < 1e-3:
return '{:.0f} \u03bcs'.format(t / 1e-6)
if t < 1:
return '{:.0f} ms'.format(t / 1e-3)
else:
return '{:.0f} s'.format(t)
def time_to_txt3(t):
if t < 1e-6:
return '{:.1f} ns'.format(t / 1e-9)
if t < 1e-3:
return '{:.1f} \u03bcs'.format(t / 1e-6)
if t < 1:
return '{:.1f} ms'.format(t / 1e-3)
else:
return '{:.1f} s'.format(t)
def count_to_txt(v):
if v <= 9999:
return '{:.0f}'.format(v)
if v <= 9999999:
return '{:.0f} K'.format(v / 1e3)
if v <= 9999999999:
return '{:.0f} M'.format(v / 1e6)
return '{:.0f} G'.format(v / 1e9)
def logn_to_txt(v):
if v == 0:
return "1"
v = 10**v
if v >= 1e9:
return '< {:.0f} G'.format(v / 1e9)
if v >= 1e6:
return '< {:.0f} M'.format(v / 1e6)
if v >= 1e3:
return '< {:.0f} K'.format(v / 1e3)
return '< {:.0f}'.format(v)
columns = []
for i, k in enumerate(sorted(j["histo_points"].keys())):
df = pd.DataFrame(j["histo_points"][k], columns=['n', 'center'])
n = df.n.sum()
total = (df.center * df.n).sum()
df["op"] = k
df["label"] = k + '\n' + count_to_txt(n) + '\n' + time_to_txt(total)
print(f"column {i:<2d}: k:{k:<15s} points:{len(df):<3d} "
f"samples:{n:<10,d} total:{total:<.9f}")
columns.append(df)
df = pd.concat(columns)
print("Stats per op -----------------")
df1 = df.copy()
df1['duration'] = df1.center * df1.n
df1 = df1.groupby("op").sum().drop(columns=['center']).reset_index()
df1['mean_duration'] = df1.duration / df1.n
df1['max_duration'] = df1.op.apply(lambda op: j["max_durations"][op][1])
df1['label'] = df.groupby("op").first().label.reset_index(drop=True)
print(df1)
print("All data points -----------------")
df["duration"] = df.center
df['logn'] = np.ceil(np.log10(df.n))
print(df)
# Seaborn plot ----------------
# https://seaborn.pydata.org/examples/scatter_bubbles.html
g = sns.relplot(
x="op",
y="duration",
size="logn",
palette="muted",
alpha=.5,
sizes=(20, 750),
height=6,
data=df,
legend="full",
# facet_kws=dict(legend_out=True, )
)
g.set(yscale="log")
plt.grid(axis='y')
v = plt.plot(df1["op"],
df1.max_duration,
'_',
color='black',
alpha=0.5,
ms=15,
label="max")
w = plt.plot(df1["op"],
df1.mean_duration,
'_',
color='red',
alpha=0.5,
ms=15,
label="mean")
for _, row in df1.iterrows():
plt.annotate(
time_to_txt3(row.mean_duration),
(row["op"], row.mean_duration),
xytext=(10, 0),
textcoords='offset pixels',
ha='left',
va='center',
size=8,
weight='bold',
)
plt.annotate(
time_to_txt3(row.max_duration),
(row["op"], row.max_duration),
xytext=(10, 0),
textcoords='offset pixels',
ha='left',
va='center',
size=8,
weight='bold',
)
plt.legend(handles=v + w, loc="upper right")
plt.xlabel('op name / call count / total time spent in op', weight="bold")
plt.ylabel('duration of call (log scale)', weight="bold")
# Legend ----------------
g.legend.set_title("Number of calls")
for txt in g.legend.texts:
print(float(txt.get_text()), logn_to_txt(float(txt.get_text())))
txt.set_text(logn_to_txt(float(txt.get_text())))
# Title ----------------
el_wall_inop = time_to_txt(df1.duration.sum())
el_wall = time_to_txt(float(j["elapsed"]))
el_cpu = time_to_txt(float(j["elapsed_cpu"]))
flatten = {
"0": False,
"1": True,
0: False,
1: True,
}[j["flatten"]]
plt.title(
("bench/tree.exe on the bootstrap trace / commits:{} / flatten:{} / inode-config:{}\n"
"wall in ops:{} / wall:{} / cpu:{} / revision:{}")
.format(
df[df["op"].str.contains("Commit")].n.sum(),
flatten,
j["inode_config"],
el_wall_inop, el_wall, el_cpu,
j["revision"],
), weight="bold")
# Tick label x ----------------
xticks, _ = plt.xticks()
plt.xticks(xticks, df1.label)
# Tick label y ----------------
a = min(df["duration"].min(), 1e-7)
b = max(pd.DataFrame(j["max_durations"]).T.iloc[:, 1].max(), 1e+3)
a, b = np.log10([a, b])
yticks = 10**np.arange(np.ceil(a), np.floor(b) + 1)
ymin = 10**(np.floor(a * 4) / 4)
ymax = 10**(np.ceil(b * 4) / 4)
plt.ylim(ymin, ymax)
plt.yticks(yticks, [time_to_txt2(v) for v in yticks])
plt.minorticks_off()
# Save ----------------
# https://stackoverflow.com/a/56970556
g.fig.set_figwidth(10)
g.fig.set_figheight(6)
g.tight_layout()
g.savefig(args.out)
Raw
moving_average_tree.py
"""Plot distribution of timings out of bench/tree.exe on boostrap trace
Formatted using YAPF
"""
import argparse
import json
import os
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib
import matplotlib.pyplot as plt
sns.set_theme(style="ticks", color_codes=True)
# plt.rcParams["legend.labelspacing"] = 1.7
p = argparse.ArgumentParser()
p.add_argument('--in',
help="""path to JSON file.
Pseudo schema:
{
"revision" : string;
"flatten": int;
"inode_config": string;
"max_durations": {string -> (int, float)};
"moving_average_points": {string -> {"x": float array; "y": float array}};
"histo_points": {string -> (int * float) array};
}
""")
p.add_argument('--out', help="Suffix of output filename")
p.add_argument('-o',
help="Overwrite if --out exists. Fail otherwise",
action="store_true")
args = p.parse_args()
print(args)
j = json.loads(open(dict(args._get_kwargs())['in']).read())
if args.o == False and os.path.exists(args.out):
raise NameError("{} exists!! Pass -o to overrite it".format(args.out))
def time_to_txt(t):
h = np.floor(t / 3600)
t -= h * 3600
m = np.floor(t / 60)
t -= m * 60
return "{:02.0f}:{:02.0f}:{:02.0f}".format(h, m, t)
def time_to_txt2(t):
if t < 1e-6:
return '{:.0f} ns'.format(t / 1e-9)
if t < 1e-3:
return '{:.0f} \u03bcs'.format(t / 1e-6)
if t < 1:
return '{:.0f} ms'.format(t / 1e-3)
else:
return '{:.0f} s'.format(t)
def time_to_txt3(t):
if t < 1e-6:
return '{:.1f}ns'.format(t / 1e-9)
if t < 1e-3:
return '{:.1f}\u03bcs'.format(t / 1e-6)
if t < 1:
return '{:.1f}ms'.format(t / 1e-3)
else:
return '{:.1f}s'.format(t)
def count_to_txt(v):
if v <= 9999:
return '{:.0f}'.format(v)
if v <= 9999999:
return '{:.0f} K'.format(v / 1e3)
if v <= 9999999999:
return '{:.0f} M'.format(v / 1e6)
return '{:.0f} G'.format(v / 1e9)
def logn_to_txt(v):
if v == 0:
return "1"
v = 10**v
if v >= 1e9:
return '< {:.0f} G'.format(v / 1e9)
if v >= 1e6:
return '< {:.0f} M'.format(v / 1e6)
if v >= 1e3:
return '< {:.0f} K'.format(v / 1e3)
return '< {:.0f}'.format(v)
columns = []
for i, k in enumerate(sorted(j["histo_points"].keys())):
df = pd.DataFrame(j["histo_points"][k], columns=['n', 'center'])
n = df.n.sum()
total = (df.center * df.n).sum()
df["op"] = k
df["label"] = k + '\n' + count_to_txt(n) + '\n' + time_to_txt(total)
print(f"column {i:<2d}: k:{k:<15s} points:{len(df):<3d} "
f"samples:{n:<10,d} total:{total:<.9f}")
columns.append(df)
print("Stats per op -----------------")
df1 = pd.concat(columns).copy()
df1['duration'] = df1.center * df1.n
df1 = df1.groupby("op").sum().drop(columns=['center']).reset_index()
df1['mean_duration'] = df1.duration / df1.n
df1['max_duration'] = df1.op.apply(lambda op: j["max_durations"][op][1])
df1['label'] = df.groupby("op").first().label.reset_index(drop=True)
print(df1)
colors = dict(
Add="brown",
Checkout="grey",
Commit="red",
Copy="pink",
Find="orange",
Mem="blue",
Mem_tree="yellow",
Remove="purple",
)
plt.figure(figsize=(10, 6))
plt.yscale("log")
ncommits = ...
columns = []
for i, k in enumerate(sorted(j["moving_average_points"].keys())):
df = pd.DataFrame(j["histo_points"][k], columns=['n', 'center'])
n = df.n.sum()
if k == "Commit": ncommits = n
df = pd.DataFrame(j["moving_average_points"][k])
if len(df) == 0: continue
c = colors.get(k, "black")
plt.plot(df["xs"], df["ys"], c=c)
maxi, maxd = j["max_durations"][k]
maxx = maxi / (n - 1)
print(k, maxi, n, maxx, maxd)
plt.plot([maxx], [maxd],
'X',
color=c,
alpha=0.5,
ms=10,
label="{}\n• {}\n• {}".format(k, time_to_txt3(maxd), maxi))
# label="{}, max at idx {}".format(k, maxi))
# plt.annotate(
# time_to_txt3(maxd),
# (maxx, maxd),
# xytext=(10, 0),
# textcoords='offset pixels',
# ha='left',
# va='center',
# size=8,
# weight='bold',
# )
columns.append(df)
df = pd.concat(columns)
print(df)
plt.xlabel('trace replay progress', weight="bold")
plt.ylabel('duration of call (log scale) (smoothed curve)', weight="bold")
plt.grid(axis='y')
# Title ----------------
el_wall_inop = time_to_txt(df1.duration.sum())
el_wall = time_to_txt(float(j["elapsed"]))
el_cpu = time_to_txt(float(j["elapsed_cpu"]))
flatten = {
"0": False,
"1": True,
0: False,
1: True,
}[j["flatten"]]
plt.title(
("bench/tree.exe on the bootstrap trace / commits:{} / flatten:{} / inode-config:{}\n"
"wall in ops:{} / wall:{} / cpu:{} / revision:{}")
.format(
ncommits,
flatten,
j["inode_config"],
el_wall_inop, el_wall, el_cpu,
j["revision"],
), weight="bold")
# Tick label x ----------------
xticks = np.linspace(0, 1, 10)
plt.xticks(xticks, [
time_to_txt(x * float(j["elapsed"])) for x in xticks
])
plt.xlim(0, 1)
# Tick label y ----------------
a = min(df["ys"].min(), 1e-7)
b = max(pd.DataFrame(j["max_durations"]).T.iloc[:, 1].max(), 1e+3)
print(a, b)
a, b = np.log10([a, b])
print(a, b)
a = np.floor(a)
b = np.ceil(b)
print(a, b)
print(10**a, 10**b)
yticks = 10**np.arange(a, b + 1)
ymin = 10**a
ymax = 10**b
plt.ylim(ymin, ymax)
plt.yticks(yticks, [time_to_txt2(v) for v in yticks])
plt.minorticks_off()
plt.legend(fontsize="small", bbox_to_anchor=(1.0, 1), loc='upper left',
title="Operation\n• len of max\n• idx of max")
# Save ----------------
# https://stackoverflow.com/a/56970556
# g.fig.set_figheight(6)
plt.tight_layout()
plt.savefig(args.out)
Raw
show_off.py
import datetime
import argparse
import json
import os
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib
import matplotlib.pyplot as plt
# import plotly.express as px
# import plotly.graph_objects as go
class BubbleChart:
def __init__(self, area, bubble_spacing=0):
"""
Setup for bubble collapse.
Parameters
----------
area : array-like
Area of the bubbles.
bubble_spacing : float, default: 0
Minimal spacing between bubbles after collapsing.
Notes
-----
If "area" is sorted, the results might look weird.
"""
area = np.asarray(area)
r = np.sqrt(area / np.pi)
self.bubble_spacing = bubble_spacing
self.bubbles = np.ones((len(area), 4))
self.bubbles[:, 2] = r
self.bubbles[:, 3] = area
self.maxstep = 2 * self.bubbles[:, 2].max() + self.bubble_spacing
self.step_dist = self.maxstep / 2
# calculate initial grid layout for bubbles
length = np.ceil(np.sqrt(len(self.bubbles)))
grid = np.arange(length) * self.maxstep
gx, gy = np.meshgrid(grid, grid)
self.bubbles[:, 0] = gx.flatten()[:len(self.bubbles)]
self.bubbles[:, 1] = gy.flatten()[:len(self.bubbles)]
self.com = self.center_of_mass()
def center_of_mass(self):
return np.average(
self.bubbles[:, :2], axis=0, weights=self.bubbles[:, 3]
)
def center_distance(self, bubble, bubbles):
return np.hypot(bubble[0] - bubbles[:, 0],
bubble[1] - bubbles[:, 1])
def outline_distance(self, bubble, bubbles):
center_distance = self.center_distance(bubble, bubbles)
return center_distance - bubble[2] - \
bubbles[:, 2] - self.bubble_spacing
def check_collisions(self, bubble, bubbles):
distance = self.outline_distance(bubble, bubbles)
return len(distance[distance < 0])
def collides_with(self, bubble, bubbles):
distance = self.outline_distance(bubble, bubbles)
idx_min = np.argmin(distance)
return idx_min if type(idx_min) == np.ndarray else [idx_min]
def collapse(self, n_iterations=50):
"""
Move bubbles to the center of mass.
Parameters
----------
n_iterations : int, default: 50
Number of moves to perform.
"""
for _i in range(n_iterations):
moves = 0
for i in range(len(self.bubbles)):
rest_bub = np.delete(self.bubbles, i, 0)
# try to move directly towards the center of mass
# direction vector from bubble to the center of mass
dir_vec = self.com - self.bubbles[i, :2]
# shorten direction vector to have length of 1
dir_vec = dir_vec / np.sqrt(dir_vec.dot(dir_vec))
# calculate new bubble position
new_point = self.bubbles[i, :2] + dir_vec * self.step_dist
new_bubble = np.append(new_point, self.bubbles[i, 2:4])
# check whether new bubble collides with other bubbles
if not self.check_collisions(new_bubble, rest_bub):
self.bubbles[i, :] = new_bubble
self.com = self.center_of_mass()
moves += 1
else:
# try to move around a bubble that you collide with
# find colliding bubble
for colliding in self.collides_with(new_bubble, rest_bub):
# calculate direction vector
dir_vec = rest_bub[colliding, :2] - self.bubbles[i, :2]
dir_vec = dir_vec / np.sqrt(dir_vec.dot(dir_vec))
# calculate orthogonal vector
orth = np.array([dir_vec[1], -dir_vec[0]])
# test which direction to go
new_point1 = (self.bubbles[i, :2] + orth *
self.step_dist)
new_point2 = (self.bubbles[i, :2] - orth *
self.step_dist)
dist1 = self.center_distance(
self.com, np.array([new_point1]))
dist2 = self.center_distance(
self.com, np.array([new_point2]))
new_point = new_point1 if dist1 < dist2 else new_point2
new_bubble = np.append(new_point, self.bubbles[i, 2:4])
if not self.check_collisions(new_bubble, rest_bub):
self.bubbles[i, :] = new_bubble
self.com = self.center_of_mass()
if moves / len(self.bubbles) < 0.1:
self.step_dist = self.step_dist / 2
def print_time_elapsed(s):
rhs = str(datetime.timedelta(seconds=s % 3600))[:-7]
lhs = s // 3600
if lhs == 0:
return rhs
else:
return f'{lhs:.0f}{rhs[1:]}'
p = argparse.ArgumentParser()
p.add_argument('--in', help="""path to JSON file""", required=True)
p.add_argument('--out', help="output path (e.g. a png or svg path)", required=True)
p.add_argument('-o',
help="Overwrite if --out exists. Fail otherwise",
action="store_true")
args = p.parse_args()
print(args)
if args.o == False and os.path.exists(args.out):
raise NameError("{} exists!! Pass -o to overrite it".format(args.out))
j = json.loads(open(dict(args._get_kwargs())['in']).read())
print(j.keys())
print(j['config'])
print(j['hostname'])
print()
print('> SPAN:')
print(j['span'].keys())
print(j['span']['mem'].keys())
print(j['span']['mem']['count'].keys())
print()
print('> GC:')
print(j['gc'].keys())
print(j['gc']['major_collections'].keys())
print(j['gc']['major_collections']['diff_per_block'].keys())
print()
print(j['span']['mem']['count']['min_value'])
print(j['span']['mem']['count']['max_value'])
print(j['span']['mem']['count']['mean'])
print(j['span']['mem']['cumu_count']['diff'])
block_count = j['block_count']
total_op_count = j['op_count']
total_wall = j['elapsed_wall']
print('total_op_count', total_op_count)
print('block_count', block_count)
print('total_wall', total_wall, total_wall / 3600)
colors = {
# https://coolors.co/palettes/trending
# # https://coolors.co/d9ed92-b5e48c-99d98c-76c893-52b69a-34a0a4-168aad-1a759f-1e6091-184e77
# 'checkout': '#d9ed92',
# 'find': '#b5e48c',
# 'mem': '#99d98c',
# 'mem_tree': '#76c893',
# 'add': '#52b69a',
# 'remove': '#34a0a4',
# 'copy': '#168aad',
# 'commit': '#1a759f',
# 'unseen': '#1e6091',
# 'buildup': '#184e77',
# https://coolors.co/03045e-023e8a-0077b6-0096c7-00b4d8-48cae4-90e0ef-ade8f4-caf0f8
'': '#03045e',
'copy': '#023e8a',
'mem': '#0077b6',
'commit': '#0096c7',
'add': '#00b4d8', 'checkout': '#00b4d8',
'remove': '#48cae4',
'find': '#90e0ef',
'mem_tree': '#ade8f4',
'unseen': '#caf0f8',
}
labels = dict(
unseen='in-between ops',
buildup='others',
)
# https://plotly.com/python/pie-charts/
# https://matplotlib.org/stable/gallery/misc/packed_bubbles.html#sphx-glr-gallery-misc-packed-bubbles-py
rows = []
for opk in sorted(j['span'].keys()):
rows.append(dict(label=labels.get(opk, opk),
key=opk,
average_duration=j['span'][opk]['duration']['mean'],
count_per_block=j['span'][opk]['count']['mean'],
explode=0.2 if opk == 'buildup' else 0,
color=colors.get(opk, 'grey')))
df0 = pd.DataFrame(rows)
# **************************************************************************************************
# **************************************************************************************************
# **************************************************************************************************
plt.close('all')
fig, ax = plt.subplots(figsize=np.asarray([10, 5]) * 0.6666)
ax2 = ax.twinx()
ax.set_yscale('log')
# ax2.set_yscale('linear')
y0 = np.asarray(j['index']['bytes_written']['value_after_commit']['evolution']).astype('float')
x = np.linspace(0, block_count - 1, y0.size)
ax.plot(x, y0, label='written so far', color=colors['mem'])
ax.annotate(
f"{y0[-1] / 1e12:.0f}TB",
[(block_count - 1) * 1.01 , y0[-1]], verticalalignment="center", horizontalalignment='left',
)
y1 = np.asarray([d['value_after_commit']['evolution'] for d in j['disk'].values()]).astype(float).sum(axis=0)
ax.plot(x, y1, label='irmin-pack', color=colors['find'])
ax.annotate(
f"{y1[-1] / 1e9:.0f}GB",
[(block_count - 1) * 1.01 , y1[-1]], verticalalignment="center", horizontalalignment='left',
)
y2 = y0 / y1
ax2.plot(x, y2, '--', label='write amplification', color='grey')
ax2.annotate(
f"x{y2[-1]:.0f}",
[(block_count - 1) * 1.01 , y2[-1]], verticalalignment="center", horizontalalignment='left',
)
ax.set_ylim(100_000_000, 10 ** (np.log10(y0.max()) * 1.025))
ax.set_yticks([
1e9,
1e10,
1e11,
1e12,
1e13,
# 1e14,
])
ax.minorticks_off()
ax2.set_ylim(1, np.nanmax(y2) * 1.2)
ax2.set_yticks([1, 100, 200, 300])
ax2.set_yticklabels(["x1", "x100", "x200", "x300"])
ax.set_yticklabels([
"1GB",
"10GB",
"100GB",
"1TB",
"10TB",
# "100TB",
])
ax.set_xticks(
[
0,
# 28_083,
204_762,
458_753,
655_361,
851_969,
1_212_417,
1_343_489,
# 1_466_367,
])
ax.set_xticklabels(
[
"genesis",
# "alpha II",
"alpha III",
"athens A",
"babylon",
"carthage",
"dephi",
'edo',
# "florence",
])
ax.set_xlim(0 - block_count * 0.05, block_count * 1.12)
ax.xaxis.grid()
# ax.grid(which='x')
ax.legend(loc='upper left')
ax2.legend(loc='lower right')
fig.tight_layout()
plt.savefig('write_amplif.png', dpi=150)
# **************************************************************************************************
# **************************************************************************************************
# **************************************************************************************************
plt.close('all')
fig, ax = plt.subplots(figsize=np.asarray([10, 4]) * 0.6666)
y = np.asarray(j['gc']['major_heap_bytes']['value_after_commit']['evolution']).astype('float') / 1e9
x = np.linspace(0, block_count - 1, y.size)
ax.plot(x, y, label='smoothed average', color=colors['copy'])
ax.annotate(
f"{y[-1]:.2f}GB",
[(block_count - 1) * 1.01 , y[-1]], verticalalignment="center", horizontalalignment='left',
)
y = np.asarray(j['gc']['major_heap_top_bytes']).astype('float') / 1e9
mask = np.r_[True, y[1:] != y[:-1]]
ax.plot([0, block_count - 1], [y.max(), y.max()], color=colors['remove'])
ax.plot([x[mask][-1]], [y[mask][-1]], 'x', label='highest recorded', color=colors['remove'])
ax.annotate(
f"{y.max():.2f}GB",
[(block_count - 1) * 1.01 , y.max()], verticalalignment="center", horizontalalignment='left',
)
ax.set_yticks([0, 1, 2, 3])
ax.set_yticklabels(
[
'0',
'1GB',
'2GB',
'3GB',
])
ax.set_ylim(0, y.max() * 1.08)
ax.set_xticks(
[
0,
# 28_083,
204_762,
458_753,
655_361,
851_969,
1_212_417,
1_343_489,
# 1_466_367,
])
ax.set_xticklabels(
[
"genesis",
# "alpha II",
"alpha III",
"athens A",
"babylon",
"carthage",
"dephi",
'edo',
# "florence",
])
ax.set_xlim(0 - (block_count / 40), block_count * 1.12)
ax.grid()
ax.legend(loc='lower right')
fig.tight_layout()
plt.savefig('memory.png', dpi=150)
# **************************************************************************************************
# **************************************************************************************************
# **************************************************************************************************
plt.close('all')
fig, ax = plt.subplots(subplot_kw=dict(aspect="equal"), figsize=(5, 5))
df = df0.set_index('key').loc['checkout find mem mem_tree add remove copy commit'.split(' ')].reset_index()
index = list(df.index)
seed = int(np.random.rand() * 1000)
seed = 642
np.random.RandomState(seed).shuffle(index)
print("seed!", seed)
df = df.loc[index]
print(df)
bubble_chart = BubbleChart(area=df['count_per_block'], bubble_spacing=4)
bubble_chart.collapse(5)
for i in range(len(bubble_chart.bubbles)):
row = df.iloc[i]
circ = plt.Circle(
bubble_chart.bubbles[i, :2], bubble_chart.bubbles[i, 2], color=row['color'])
ax.add_patch(circ)
text = row['label'] + '\n{:.1f}'.format(row['count_per_block'])
if row['label'] in 'commit remove checkout copy mem_tree'.split(' '):
ax.text(bubble_chart.bubbles[i, 0] + bubble_chart.bubbles[i, 2] / 2 * 1.8 + 1,
bubble_chart.bubbles[i, 1],
text,
horizontalalignment='left', verticalalignment='center',
)
else:
ax.text(*bubble_chart.bubbles[i, :2], text,
horizontalalignment='center', verticalalignment='center')
ax.axis("off")
ax.relim()
ax.autoscale_view()
# ax.set_title('Average Operation Count per Block')
plt.tight_layout()
plt.savefig('counts.png', dpi=100)
# **************************************************************************************************
# **************************************************************************************************
# **************************************************************************************************
plt.close('all')
fig, [ax0, ax1] = plt.subplots(ncols=2, subplot_kw=dict(aspect="equal"), figsize=np.asarray([10, 5]) * 0.75)
df = df0.set_index('key').loc['buildup commit unseen'.split(' ')].reset_index()
print(df)
def mypct(v):
return f'{v:.1f}%'
ax0.set_title('All')
wedges, labs, pcts = ax0.pie(df.average_duration,
labels=df.label,
explode=df['explode'],
autopct=mypct,
pctdistance=0.76,
# shadow=True,
# textprops=dict(fontsize=10),
rotatelabels=True,
labeldistance=1.03,
startangle=360 / df['average_duration'].sum() * df.set_index('key').loc['buildup', 'average_duration'] / -2,
colors=df['color'],
normalize=True,
)
for t in pcts:
t.set_fontsize(9)
df = df0.set_index('key').loc['copy find mem checkout mem_tree add remove'.split(' ')].reset_index()
print(df)
def mypct(v):
v *= df0.set_index('key').loc['buildup'].average_duration / df0.set_index('key').loc['block'].average_duration
return f'{v:.2f}%'
ax1.set_title('Others')
wedges, labs, pcts = ax1.pie(df.average_duration,
labels=df.label,
explode=df['explode'],
autopct=mypct,
pctdistance=0.76,
# shadow=True,
rotatelabels=True,
labeldistance=1.03,
startangle=360 / df['average_duration'].sum() * df.set_index('key').loc['copy', 'average_duration'] / -2 + 180,
colors=df['color'],
normalize=True,
)
for t in pcts:
t.set_fontsize(9)
plt.tight_layout()
plt.savefig('durations.png', dpi=150)
Display the source blob
Display the rendered blob
Raw
tree.ipynb
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment